Make the accepted log grep pattern more strict in order to not pick up any log
entries from sshd causing intermittent failures.

Discussed with bluhm@ back in August this year.

In the sendsyslog-dropped test the client has to write a log message
before syslogd is running. To fix a race in regress, delay the
syslogd startup until the message has been written. For debugging
add the process class to all die messages.

Check that the hostname added by syslogd -h in file and server log
is correct. Use generic redo functions. Do syntax check for all
Perl files.

Syslogd's debug and error messages have changed a bit, adapt test.

The TCP socket buffer size for syslogd has changed. Adapt tests
where we count the dropped messages. A different number gets stuck
in the kernel buffers now which is not included in the syslogd
statistics.

Add test for sendsyslog(2) dropped message reporting.

Make syslogd tests more reliable. Allow TLS tests to drop a wider
rage of messages and make dropped regexes more specific.

Test octet counting and non transparent framing.

Do not count the dropped messages strictly, so that the syslogd
tests are depending less on the timing. Fix a race in the buffer
sync test. Now make regress of syslogd passes on my amd64, i386,
sparc64, alpha machines.

Adapt the maxunix test, the syslogd -a error semantic has changed.
Increase timeouts and remove races, so that the syslogd tests pass
on my sparc64 machine.

Make syslogd tests more reliable. Increase the allowed range of
lost log messages. Improve the synchronisation of client and server.

Make syslogd test more reliable. Use Time::HiRes to sleep .01
seconds in the write_lines() loop. This slows down the client, so
all messages reach syslogd and flood its buffers. Increase the
timeout to wait until all buffers are full. Use this code in all
TCP/TLS buffer and memory overflow tests.

Change syslog tests from transparent framing to octet counting.
Add tests for truncated buffer handling.

In the sendsyslog-dropped test the client has to write a log message
before syslogd is running. To fix a race in regress, delay the
syslogd startup until the message has been written. For debugging
add the process class to all die messages.

Check that the hostname added by syslogd -h in file and server log
is correct. Use generic redo functions. Do syntax check for all
Perl files.

Syslogd's debug and error messages have changed a bit, adapt test.

The TCP socket buffer size for syslogd has changed. Adapt tests
where we count the dropped messages. A different number gets stuck
in the kernel buffers now which is not included in the syslogd
statistics.

Add test for sendsyslog(2) dropped message reporting.

Make syslogd tests more reliable. Allow TLS tests to drop a wider
rage of messages and make dropped regexes more specific.

Test octet counting and non transparent framing.

Do not count the dropped messages strictly, so that the syslogd
tests are depending less on the timing. Fix a race in the buffer
sync test. Now make regress of syslogd passes on my amd64, i386,
sparc64, alpha machines.

Adapt the maxunix test, the syslogd -a error semantic has changed.
Increase timeouts and remove races, so that the syslogd tests pass
on my sparc64 machine.

Make syslogd tests more reliable. Increase the allowed range of
lost log messages. Improve the synchronisation of client and server.

Make syslogd test more reliable. Use Time::HiRes to sleep .01
seconds in the write_lines() loop. This slows down the client, so
all messages reach syslogd and flood its buffers. Increase the
timeout to wait until all buffers are full. Use this code in all
TCP/TLS buffer and memory overflow tests.

Change syslog tests from transparent framing to octet counting.
Add tests for truncated buffer handling.